On 08/23/2012 02:44 PM, Paul Eggert wrote: > I worry that this may break on older GCCs that > don't let you put attributes at the start of a declaration.
Ah, I now see that we have such declarations elsewhere, so if this were a problem we'd probably have run into it by now. So please don't bother looking for old GCCs that reject __attribute__ at the start of a declaration. I'll look into fixing the other "void _Noreturn"s that we have in gnulib.
